NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCQV53CCHZ05GAKD1SSCD1T
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
24.0s
Cost:
$0.031040
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0001%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% words of dialogue between two thieves named Molly and Bill who are planning to break into a children's party to steal all the presents. Only include the prose of the scene without any markdown or commentary.
Model response
Messages
Meta
AI
The van smelled of cold sausage rolls and ambition. Molly had spread the blueprint across the dashboard, which was in fact a paper placemat from the Happy Otter Family Restaurant, annotated in crayon. "Right," she said. "Ground floor. Bouncy castle here, trestle table of presents here, and a clown." "A clown." Bill lowered his binoculars. "You didn't mention a clown." "I'm mentioning him now." "There's a hierarchy to information, Molly. Clown goes near the top." "He's forty-eight and his name is Trevor and he does eleven parties a weekend. By half three he'll have lost the will to live. He is not the obstacle." "What's the obstacle?" "The nan." Molly tapped the placemat, right on a ketchup smear. "Nan sits here. Nan sees everything. Nan has been watching people take things off tables since 1961." Bill wrote *NAN* in his notebook and underlined it twice, then a third time, which felt excessive but honest. "So we go in during the cake," he said. "We go in during the cake. Lights off, twenty-three children screaming, everybody staring at a burning number six. That's our window. Forty seconds." "Forty's tight." "Forty's generous. You've done banks in twenty." "Banks don't have party bags, Molly. Banks have a floor plan. Children distribute themselves like spilled marbles." He peered through the binoculars again. "There's one in the hedge." "Ignore the one in the hedge." "He's been in the hedge eleven minutes. He's either hiding or he lives there." Molly took the binoculars, looked, and handed them back. "That's Callum. Callum is not our concern. Our concern is that the presents are wrapped, which means we don't know what we're stealing, which means we take all of it and sort it in the van like professionals." "Professionals," Bill repeated, warmly, as though she'd said his name. "You take the far end. Big boxes first, they're the bikes and the scooters. I take the envelopes, because envelopes are cash and cash is God." "And the trampoline?" "We are not stealing a trampoline." "It's a *good* trampoline." "Bill." She turned to face him fully. "We have discussed this. We discussed it at length, in the kitchen, at two in the morning, while you drew a trampoline on the back of an electricity bill. It does not fold. It does not fit. It is not *in scope*." Bill looked out at the garden for a long moment, at the trampoline glowing in the afternoon light like something out of scripture. "Fine," he said. "Say it." "It's not in scope." "Good boy. Now — costumes." She reached behind the seat and produced two shapeless mounds of purple fun fur. Bill stared. "What is that?" "Entertainment. We're the Wiggly Wonder Bears. Booked, invoiced, expected at four. Nobody questions a bear at a birthday, Bill, it's basic psychology." "It's got a zip up the front." "It's got a zip up the *back*, which is why we do each other's, which is why we practise now and not on the lawn." She thrust a paw at him. "Also Trevor may want to fight us. Territorial thing. Clowns and bears." Bill took the costume. Somewhere in the garden, a child began to shriek with joy, and the operation began.
created_at
"2026-07-25T13:36:11.892Z"
7/25/2026, 1:36:11 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3809
moderation_latency
(null)
generation_time
23892
tokens_prompt
75
tokens_completion
846
native_tokens_prompt
93
native_tokens_completion
1223
native_tokens_completion_images
(null)
native_tokens_reasoning
45
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"end_turn"
service_tier
"default"
usage
0.03104
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNs7p1UydMVtKtCLpBe2"
is_byok
false
latency
1467
model_permaslug
"anthropic/claude-opus-5-20260723"
provider_name
"Amazon Bedrock"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1784986571-6CWkMX7Oypft5PxhXr2W"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784986571-BxUfEFk6Ab7aVXfRuqhK"
upstream_id
"msg_011CdNs7p1UydMVtKtCLpBe2"
total_cost
0.03104
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
534
0.00%
Dialogue to Total Word Ratio
Ratio: 69.22%, Deviation: 19.22%
neededClean
false
wordsTotal
536
wordsDialogue
371
0.0001%